基于进化策略的函数优化问题研究

Function Optimization Problems Based on Evolutionary Strategies

针对进化策略算法在解决具体问题是熟练速度较慢这一问题的原因进行分析,提出自适应变异步长的方法,以在全局和局部范围内进行搜索.变异步长的值依赖于目标变量与全局最优解之间的距离.步长随距离自适应变化,可避免局部熟练和早熟.通过对经典dejong函数和Shubert函数的仿真试验,验证了文中算法的有效性.仿真结果表明,该算法收敛速度快,搜索精度高,且具有良好的全局搜索能力.

Based on the reason of the computed speed slow for the complicate problem using evolutionary strategies, the paper presents a new adaptive mutation method. This algorithm employ self- adaptive evolutionary programming to control the population＇s global evolve. The self- adaptive evolutionary programming depends on the distance between the target variable and the globally optimal solution, and uses a mutation operator to search the solution space. The mutation operator based local optimization method and random restart method was combined in order to increase the reliability of finding the global optimum. The new algorithm can obtain satisfactory results in limited time.